The Governance Gap: Why Agentic AI Is Outrunning Internal Controls

Most finance and GBS leaders are deploying AI agents first and figuring out governance later, and that sequence is the problem. Once an agent is embedded in journal entry review, or exception handling, retrofitting controls means untangling live workflows, re-establishing audit trails after the fact, and explaining to auditors why oversight arrived after autonomy did.

This session makes the case for the opposite approach: treating internal controls as a design input for your AI strategy, not a compliance afterthought bolted on once something goes wrong. We'll look at what's actually working in finance organizations that got this sequencing right, where SOX and audit expectations are colliding with Agentic deployment speed, and how GBS teams can build the accountability structure before the agent goes live, not after.

What you'll take away:

• Why "deploy fast, govern later" creates technical debt you can't audit your way out of
• The four governance architectures finance leaders are actually using and which fits which risk profile
• The importance of harness engineering
• Where Agentic AI crosses the line from supporting a control to being the control
• How to define materiality and risk thresholds for agent autonomy before you scale
• What "audit-ready by default" looks like in practice: action logging, escalation paths, and human override built into the workflow, not added on top of it
• A practical starting sequence for GBS teams: what to lock down before pilot, not after production
  

---

Speakers

Sponsor

---

Register